Contents - Previous - Next

Effects of macrobiotic diets on linear growth in infants and children until 10 years of age

1. Introduction
2. Subjects and methods
3. Conversion into nutrients
4. Statistical analysis
5. Results
6. Recommendations for the macrobiotic diet
7. Two-year follow-up study
8. Discussion
Discussion of papers by Allen, Neumann & Harrison and Dagnelie et al.

P.C. Dagnelie, M. van Dusseldorp, W.A. van Staveren and J.G.A.J. Hautvast

Department of Human Nutrition, Agricultural University Wageningen, PO. Box 8129, 6700 EV Wageningen, The Netherlands

Correspondence to: M. van Dusseldorp.

In order to study the relationship between diet and growth, we have assessed growth patterns in children 0-10 years old on macrobiotic diets. A cross-sectional anthropometric study (0-8 years old children, n = 243) indicated that deviation from the reference growth curve occurred during the weaning period. Between 2 and 4 years there was a partial catch-up for weight and arm circumference but not for height. As a next step, a mixed-longitudinal study was performed in 4-18-month-old macrobiotic infants (n = 53) and matched omnivorous controls (n = 57). For a period of 6 months, data on growth and dietary intake were collected. The data on linear growth supported the findings on growth stagnation observed cross-sectionally. Linear growth was associated with the protein content of the diet, but not with energy intake. On the basis of our findings nutritional modifications to the macrobiotic diet (addition of fat and fish) were recommended for all macrobiotic families. Six months later (two years after the first cross-sectional data collection) the anthropometric study was repeated in the same cohort (n = 194). This follow-up study revealed that children from families which, since the initial study, had increased the consumption of fatty fish, dairy products, or both, had grown in height more rapidly than the remaining children (P < 0.05). Since no indications were found for the presence of adverse social circumstances, infectious diseases or other confounding factors, our data clearly demonstrate that linear growth retardation in children on macrobiotic diets is caused by nutritional deficiencies alone.

1. Introduction

The workshop of the International Dietary Energy Consultancy Group in Guatemala (1987) on chronic energy deficiency came to the conclusion that more studies are needed on the relationship between the intakes of energy and nutrients and the pattern of growth: "Such studies are particularly illuminating if they compare groups with different dietary patterns and are particularly important for children below three years" (Schürch & Scrimshaw, 1987). From 1985 onwards we have studied growth patterns of children on macrobiotic diets in the Netherlands, because these children appeared to be significantly lighter and shorter than a control group fed an omnivorous diet (Van Staveren et al., 1985). The macrobiotic diet has characteristics similar to the diet of many children in developing countries, in that it is also mainly composed of foods high in starch and fibre but low in protein. In developing countries, however, such a type of diet often coincides with other unfavorable circumstances which may influence linear growth. In contrast, in our studies we did not find any evidence for the existence of adverse social circumstances, infectious diseases or other confounding factors. Thus, this group of macrobiotic children enables us to study the effects of diet alone.

Macrobiotic children have a very restricted, almost vegan type of diet consisting of grain cereals (mainly rice), vegetables, pulses and sea vegetables, with only small amounts of cooked fruits and occasionally some fish. Meat, dairy products and vitamin D supplements are not being used and fish is rarely given to young children. Intakes of calcium, riboflavin and vitamin D recorded for the macrobiotically fed children were substantially below the Dutch RDAs (Netherlands Nutrition Council, 1981). Our aim was to answer the following questions:

1. At what age does growth faltering in height occur in children fed macrobiotic diets?
2. Which nutritional factors are associated with linear growth retardation'?
3. What is the effect of modification of the diet on linear growth?

In this report, we will speak of growth retardation when a significant difference (P < 0.01) from the median of the Dutch cross-sectional growth reference or from the growth curve of a matched omnivorous control group is observed. Catch-up growth is defined as a positive shift towards the median of the Dutch reference curve or towards the median of the curve of the omnivorous control group.

2. Subjects and methods

2.1. Subjects

Children fed macrobiotic diets: Participants were recruited through macrobiotic teachers and during the study by families already participating in the study. Eighty percent (= 173 families; 307 children) of all families within the Netherlands that were eligible for the study were included. Eligible for the study were children below eight years of age in the Netherlands who were Caucasian, had a birth weight of at least 2500 g, a gestational age of 38 weeks or more, did not suffer from a congenital disease and who had been on a macrobiotic diet as described by Kushi (1987) from birth onwards. The educational level of the parents was high: 64% of the fathers and 45% of the mothers had completed college or university education as compared to 17% for men and 9% for women of the Dutch population aged 20-49 years (Central Bureau of Statistics, 1983). Ninety-four percent of the parents had followed a macrobiotic diet for more than 2 years and 72% for more than 5 years. Participants of different studies (outlined below) were representative subgroups of this macrobiotic population.

Children fed omnivorous diets (mixed-longitudinal study): Infants were recruited through eight child health clinics throughout the Netherlands.

2.2. Design

2.2.1. Cross-sectional study In order to identify age groups most at risk for growth faltering, a cross-sectional anthropometric study was carried out in the Dutch macrobiotic child population, aged 0-8 years (Oct-Dec 1985). From 243 children the following anthropometric measurements were taken: weight, recumbent length of children below 2 years of age, standing height of children of 2 years and older, and mid-upper arm circumference. For length and height, two observers performed the measurements. Food habits were checked by means of a structured questionnaire including foods and cooking styles which were either typical or atypical for the macrobiotic diet. For detailed description of the methods see Dagnelie et al. (1988).

2.2.2. Mixed-longitudinal study As a next step a mixed-longitudinal study (Jan-Nov 1986) was carried out in macrobiotic infants 4-18 months old (n = 53, birth cohort 1985) and omnivorous controls (n = 57) matched for month of birth, sex, parity, educational level of the father and the residential area. Three cohorts of infants were monitored from 4 to 10 months, 8 to 14 months, and 12 to 18 months, respectively. Weight and recumbent length were taken at 2-monthly intervals. Two food record periods of three days each (precise weighing method) were included in the study design, just before or after the second and third anthropometric visit. A single blood sample was collected by a paediatrician within three months of completing the six-month longitudinal study. For detailed description of methods see Dagnelie et al. (1989a,b).

2.2.3. Two-year follow-up of initial study Based on the findings of the mixed-longitudinal study, dietary recommendations were given to all macrobiotic families in May-June 1987. Five to six months later (Oct-Dec 1987) the initial cross-sectional anthropometric study (Oct-Dec 1985) was repeated in the same macrobiotic population (children now aged 2-10 years, n = 194, i.e. 80% of those children participating in the initial cross-sectional study). Food habits and anthropometric measurements were obtained as in the initial study.

3. Conversion into nutrients

Food intake data were converted into nutrients using the computerized Dutch Food Composition Table (Stichting NEVO, 1986) completed with analyses from our own laboratory of 50 foods commonly used by the macrobiotic population (Willems et al., 1987).

4. Statistical analysis

Standard deviation scores (SDS) were calculated from the median (P50) and standard deviation of the Dutch reference data (Roede & Van Wieringen, 1985; Gerver, 1988). Growth velocities were either calculated as the increments from the first to the last measurement (expressed in units/year) or expressed in terms of change in standard deviation scores. Student's t-test was used to compare the means of two independent samples. Longitudinal changes in SDS were tested by paired t-test. In the mixed-longitudinal study, the level of successive measurements in all cohorts were simultaneously compared by multivariate analysis of variance (MANOVA). Multiple regression models were fitted in order to determine the independent effects of different variables on birth weight, weight gain, linear growth, and the presence of skin and muscle wasting and rickets after adjustment for confounding variables.

5. Results

5.1. Cross-sectional study in 0-8-year-old children

Birth weight: 4.3% of macrobiotic children had reported birth weights below 2500 g compared to 2% in a comparable Dutch population (P < 0.02). The average birth weight was 3360 g for boys (Dutch median 3500 g; significant difference at P < 0.001) and 3250 for girls (Dutch median: 3390 g; P < 0.001).

Growth: The curve of length/height for age (Fig. 1 shows data of girls only, curves for boys were similar) followed the median of the standard until the age of 6-8 months, after which a marked decline was observed, reaching a minimum level (P10) between 1.5 and 2 years. No catch-up was observed at older ages. The curves of weight for age and arm circumference for age showed the same pattern, but now a partial return towards the P50 of the reference was observed after 2 years of age. The curve of weight for height followed the median of the Dutch reference. These findings were similar to those of earlier studies in smaller samples (Dwyer et al., 1980; Shull et al., 1977).

Relation with diet: Birth weight showed strong positive relationships with both the frequency of consumption of dairy products and of fish by the family. SDS of weight, height and arm circumference were significantly higher in children from families consuming dairy products at least three times a week compared to children from families who rarely or never used dairy products. This association was partly, but not completely, attributed to the difference in birth weight in children from families with or without regular consumption of dairy products. No association was observed between SDS and family consumption of fish, meat or eggs (Dagnelie et al., 1988).

Since the deviation from the normal growth curve occurred during the weaning period we decided to carry out a mixed-longitudinal cohort study on macrobiotic infants between 4 and 18 months of age.

Fig. 1. Height for age of girls on macrobiotic diets in the Netherlands. - - P10, P50 and P90 of Dutch reference population (Roede & Van Wieringen, 1985).

5.2. Mixed-longitudinal study in 4-18-month-old infants

Weaning diet: Macrobiotic mothers continued to breastfeed for an average of 13.6 months (control group: 6.6 months). Complementary feeding in macrobiotic infants started at a mean age of 4.8 months with water-based sieved porridges from whole-grain cereals, followed by vegetables (at 5.7 months), sesame seed (6.4 months) and pulses (8.2 months). Fruits were rarely given and animal products were avoided by most families. In the control group, complementary feeding started with fruits at a mean age of 2.7 months, followed in the next two months by vegetables and cereals (Dagnelie et al., 1989a).

Table 1 shows that for all age groups combined, the daily intake of energy and nutrients in the three cohorts of the macrobiotic group differed significantly from that of the control group. In the macrobiotic group fat intake decreased from 37% energy at an age of 6-8 months to 17% at 14-16 months, due to the fact that fat from breastmilk was not replaced by other fat sources during weaning. With the decreasing amount of breastmilk, animal protein intake decreased from 7 g/d at 6-8 months to 2 g/d at 14-16 months, as compared to an increase from 18 to 28 g/d in the control group. The intake of calcium, riboflavin and vitamin B12 by macrobiotic infants was considerably below that of the control group (Dagnelie et al., 1989a).

Table 1. Intake of energy and nutrients by infants (6-16 months of age) on macrobiotic and omnivorous (control) diets

Macrobiotic group (n = 49) Mean ±SD

Control group (n = 57) Mean ±SD

Energy (MJ)

3.0 ±0.5

3.6 ±0.7 a

En. density (kJ/g)

2.4 ±0.3

3.4 ±0.5 a


Animal b (g)

4 ±3

24 ±8 a

Total (g)

20 ±7

32 ±10 a

%Energy (%)

11 ±3

15 ±3 a

Fat (g)

22 ±9

30 ±7 a

%Energy (%)

28 ±12

32 ±7 a


Oligosacchar. (g)

47 ±20

70 ±17 a

Polysacchar. (g)

63 ±41

45 ±18 a

Total (g)

110 ±29

115 ±25

%Energy (%)

61 ±10

54 ±5 a

Dietary fibre (g)

13 ±7

7 ±3 a

Dietary fibre (g/MJ)

4.1 ±1.9

2.0 ±0.7 a

Calcium c (mg)

280 ±68

751 ±230 a

Iron (mg)

5.1 ±2.8

4.0 ±1.6 a

Thiamine (mg)

0.6 ±0.3

0.4 ±0.1 a

Riboflavin (mg)

0.4 ±0.1

1.1 ±0.3 a

Vitamin B12 (pa)

0.3 ±0.2

2.9 ±1.3 a

Vitamin C (mg)

53 ±22

77 ±40 a

a Difference significant at P < 0.001.
b Including breast milk.
c Based on preparation of foods with water containing 28 mg of calcium per liter.

Growth: Growth curves of length and weight for age for both groups are presented in Fig. 2. Deviation in linear growth from the P50 of the reference occurred in the macrobiotic infants from four months onwards. From 16 months of age linear growth stabilized at the 10th percentile of the Dutch references, whereas in the control group growth velocities were similar to the Dutch references (Dagnelie et al., 1989b). Comparison with birth weight data of the macrobiotic infants showed that retardation of weight growth already occurred before the age of 4 months. Weight growth was most depressed between 8 and 14 months, while stabilization at the P3 level occurred between 14 and 18 months. As shown in Table 2 for the combined cohorts, growth velocities for weight, length and arm circumference (expressed as units per year) were significantly lower in the macrobiotic group.

Relation with diet: A multiple regression analysis was carried out to determine whether the observed growth retardation could be explained by a reduced intake of energy or protein by the macrobiotic infants. Both the energy intake and the protein content of the macrobiotic diet contributed independently to growth in weight and arm circumference (Dagnelie et al., 1989b). In contrast, growth in length was only associated with the protein content of the diet.

Clinical and biochemical observations (methods of blood analysis) are described in Dagnelie et al. (1989a,c). The low riboflavin intake by the macrobiotic children was reflected by an elevated activity coefficient of erythrocyte glutathion reductase (EGR) (Dagnelie et al., 1989a). Iron deficiency was observed in 15% of the macrobiotic infants, versus no infants in the control group. As a consequence of the extremely low intake of vitamin B12 in macrobiotic infants, we found low plasma vitamin B12 concentrations which were associated with a rise in mean corpuscular volume (MCV) (Dagnelie et al., 1989c). In summer, 28% of macrobiotic children showed clinical symptoms of rickets, and in winter 55%. Our data indicated that the high prevalence of rickets is the result of long-term depletion of body calcium stores caused by a diet with a low calcium and a high fiber content, in combination with vitamin D deficiency during a part of the year. No indication of an effect of the duration of breastfeeding on the presence of rickets was found (Dagnelie et al., 1990). Major skin and muscle wasting was present in 30% of the macrobiotic infants. Growth velocities in weight and height were lower in these wasted infants (P < 0.05). Infants with major skin and muscle wasting were slower in locomotor development than the other macrobiotic infants (P = 0.05) (Dagnelie et al., 1989b). The macrobiotic group was significantly later in gross motor development (P < 0.001) and, to a lesser degree, in speech and language development (P < 0.03) (Dagnelie et al., 1989b).

Fig. 2. Growth curves (length for age, weight for age) of infants on macrobiotic or on omnivorous (control) diets in the Netherlands. - - - P10, P50 and P90 of Dutch reference population (Roede & Van Wieringen, 1985).

Figure (2a)

Figure (2b)

Table 2. Growth velocities (units per year) of infants (4-18 months of age) on macrobiotic and omnivorous (control) diets

Macrobiotic group (n = 52) Mean ±SD

Control group (n = 57) Mean ±SD

Weight (kg)

3.1 ±1.6

4.4 ±1.4 a

Length (cm)

13.2 ±3.9

16.7 ±3.1 a

Arm circumference (cm)

1.0 ±2.2

2.3 ±2.4 b

Weight-for-length (kg)

2.4 ±1.9

3.3 ±1.8 c

Differences between groups significant at a P < 0.001, b P <0.003, c P < 0.02.

6. Recommendations for the macrobiotic diet

Based on the findings of the above studies the following dietary recommendations were given to all macrobiotic families in the Netherlands:

1. To add fat as an additional source of energy to a total of at least 25-30% of energy as fat. This can be achieved by including 20-25 g of oil per day, or by eating more nuts and seeds.

2. To include fish (2-3 portions per week) as a source of vitamin B12. Especially during winter fatty fish should be consumed to provide the child with vitamin D.

3. To increase the consumption of plant sources of calcium and vitamin B2. Consumption of these products is within the macrobiotic philosophy, but they are poor sources of calcium. At this stage the use of dairy products was not yet explicitly recommended since this was considered incompatible with the macrobiotic philosophy. The effects of these dietary recommendations have been evaluated in the 2-year follow-up study.


Contents - Previous - Next